The present invention relates to a non-linear distortion compensating technique for reducing non-linear distortion components output from a transmitter that transmits radio waves amplified in power by a power amplifier.
An example of a conventional non-linear distortion compensating technique, in particular, a predistorting compensation method, is described in Eijogakugiho Vol. 24, No. 79, BCS2000-92 “An Adaptive Predistortion Method for Linear Power Amplifiers” (Document 1). The following describes this technique with reference to FIG. 7 showing a block diagram of a transmitter including a prior art power amplifier.
In FIG. 7, an input signal to be transmitted goes through a distributor 51, a delay element 52, a phase controller 53 and a gain controller 54 and is power-amplified by a power amplifier 55 before output as an output signal through a directional coupler 56. Meanwhile, part of the input signal is distributed by the distributor 51 to a detector 57 and then converted to a digital signal by an A/D converter 58.
Further, part of the power-amplified output signal is distributed to a mixer 59 and converted to an intermediate frequency by a synthesizer 60. Then, out-of-band distortion power introduced by the power amplifier 55 is extracted by a BPF 61 to a detector 62 and converted by an A/D converter 68 to a digital signal.
The following describes how the non-linear distortion compensating operation is controlled. The operations of the phase controller 53 and gain controller 54 are controlled according to signals obtained by respectively converting contents recorded in tables 66 and 67 to analog signals. The distortion that is equal in amplitude but opposite in phase to the distortion introduced by the power amplifier 55 is generated by the phase controller 53 and gain controller 54 so that non-linear compensation is performed by compensating for the distortion introduced by the power amplifier 55.
In the tables 64 and 65, envelope signals detected by the detector 57 and brought in by the A/D converter 58 are given addresses. In addition, learning is performed in the computing unit 63 by the perturbation method so as to make smaller the distortion power to be detected by the detector 62. Contents in the tables 64 and 65 are successively updated to optimum values according to the learning results so as to minimize the distortion.
According to a table update method described in the aforementioned Document 1, each of the table addresses incremented in steps of 1 is given a value by the perturbation method. However, this method is very impractical since the update requires considerably long time. It is more practical to apply the perturbation method to a predetermined representative points, for example, eight representative points, instead of every point.
FIG. 8 shows how the representative points are related with the table addresses. It is assumed here that there are table addresses 1 through 1024. The addresses 1 through 1024 are represented by eight points. In FIG. 8, each of the eight points is given a table value as indicated by a black circle. The eight representative addresses are given values (heights of the black circles).
By increasing (as indicated by an upward arrow in the figure) and decreasing (as indicated by an downward arrow in the figure) a value while checking the distortion power, the value is updated to a value which decreased the distortion. This operation is repeated for the other representative points to optimize the table values. The values given to the remaining addresses other than the eight representative addresses are updated to those determined by means of interpolation using FIR filters.
This update control is performed by applying the perturbation method to the values given to eight representative points per table or a total of sixteen points while checking the distortion power so as to optimize the tables.
Another conventional technique is also described in Japanese Patent Laid-open No. 2001-168774 (Document 2). In this method, digital base band signals are extracted from the RF input and RF output of a RF amplifier to detect and eliminate the time and phase differences between the two signals.
In order to compensate for distortion components, the amount of amplitude compensation and the amount of phase compensation are selected for the detected amplitude and phase errors between the two signals from an initially registered and adaptively updated table of compensation amounts associated with amplitude values and added successively to the RF input digital base band signal.